Observed and projected changes in extreme drought and wet-prone regions over India under CMIP5 RCP8.5 using a new vulnerability index


 Past versions of vulnerability index have shown ability to detect susceptible region by assessing socio-economic parameters at local scales. However, due to variability of these vulnerability index respect to socio-economic parameters, cann’t be utilized to predict the susceptibility region. The present endeavor aims to develops a new vulnerable index which identify and predict the spatio-temporal imprint of extreme drought and wet events at various scales 1o×1o in India by analyzing monthly observed and Coupled Model Inter-Comparison Phase 5 (CMIP5) rainfall data at spatial scale of time period pertaining to 1901-2100. New vulnerability index is proposed by consolidating the outcomes of Standard Precipitation Index (SPI) at different time scales such as 3- and 12-month and along with weights of individual grids. The weights of individual grid is calculated through the occurrence of extreme drought and wet events in the recent past which is to include a climate change factor in the proposed index. Based on the spatial distribution of high index values, the expected vulnerable regions concerning extreme drought events will be in Northeast, Northeast Central, East Coast, West, Northwest, Northcentral, and some grids in South part of India. Similarly, vulnerable regions concerning extreme wet events are likely to be in the Northeast, West Coast, East Coast, and some grids in the Peninsular region.Further, a conceptual model is presented to quantify the severity of extreme events. The analyses reveal that on the CMIP5 model data, it is obtained that 2024, 2026-27, 2035, 2036-37, 2043-44, 2059-60, 2094 are likely to be the most prominent drought years in all-India monsoon rainfall and their impact will persist for a longer time. Similarly, the most prominent wet events are predicted to be 2076, 2079-80, 2085, 2090, 2092, and 2099.

degree of closeness between the two datasets. Further, gridded data is used to calculate the all-India rainfall by taking the area-weighted 158 average over all grids using a standard weighted matrix (Rajeevan et al. 2006) provided by the 159 IMD. 160 The observed data is partitioned into three parts such as     Table 1. Further, an ensemble of these selected models is formed by implementing Step 1: Shannon's entropy (Shannon 1948) is a method to find the desired weights for the given 181 criteria which can be assessed as where, is the normalized form of the decision matrix, is the number of models and is the 185 Shannon entropy, which gives the information of the individual model's weightage used in the 186 analysis.

187
Step 2: The weights can be written as Step 3: An ensemble of data is formed by probability density function of the gamma distribution is defined by: where > 0 is a shape parameter, > 0 is a scale parameter, and > 0 is the amount of 223 precipitation. ( ) is the gamma function which is defined as: The mathematical form of the cumulative probability function of Gamma distribution is given by: Since the gamma distribution is undefined for = 0, the cumulative distribution function for 228 gamma distribution which accounts zero value in the data is further modified as: where represents the probability of zero rainfall over the period 1901 − 2014.

231
The algorithm of SPI is implemented on all-India and gridded monthly rainfall data for the period  duration is assigned more weights and vice versa. The magnitude of either event is calculated as:

251
Step 1: The implementation of SPI algorithm at a given time scale on rainfall time series ( × 1),

252
where represents the total number of months, it results in SPI index, named as ×1 .

253
Transform this matrix to × , where = 12 = 114 and represents number of years and months, 254 respectively. In this study, extreme rainfall events are considered and the value of threshold ( ℎ ) 255 is ±2 for extreme wet and drought, respectively.

259
Step 3: For ℎ extreme wet year, find the ℎ month which exceeds the ℎ . Then wet magnitude to compare at a common scale.  273 Generally, the vulnerability is a relative measure among regions/grids concerning rainfall of global warming is assigned a relatively higher weight.

291
The term is defined as:  The term is defined below:   Table   334 2. The results reveal that most extreme wet events at a 3 and 12-month scale occurred in 1920-    1916-17, 1932, 1946 1955, 1970, 1977, 1989, 1997, 2005, 2009 Table A1 and Table A2. It reveals that the gamma distribution performs 642 uniformly over all kinds of datasets. A rank is assigned to each of distribution based on the lowest rank, represents best fit for the rainfall data.